Index of every page on this site: llms.txt
All skills
Run it yourself

Scans a repo's GitHub Actions workflows for the ten critical CI/CD attack vectors — template injection, fork code executed with privileges (pwn requests), cache poisoning, impostor action SHAs, secrets dumps, GITHUB_ENV hijack, write-token untrusted triggers, credentials in caches/artifacts, unverified remote code execution (curl|bash and mutable fetch-and-run), and dependency install scripts running in a job that holds secrets — reports every finding with a plain-English attacker scenario, plus pass/fail config hygiene checks, and fixes selected findings via per-finding subagents. Deliberately NOT comprehensive — critical exploit-chain checks only (references/why-these-ten.md). Use when: the user asks to audit or review CI/CD or GitHub Actions security posture, "is my CI secure", "audit my CI security", or names ci-secure / /ci-secure. Do NOT trigger for: CI speed, cost, or wall-clock audits ("why is CI slow") — use ci-speedup; CI config best-practices grading ("grade my CI", "CI score") — use ci-score.

Use this Skill: https://skilld.dev/gh/starslingdev/skills/ci-secure

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencestroubleshooting.md

≈1.1k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Common issues

Operator-facing failure modes for a ci-secure run. The behavioral rules these point back to live in SKILL.md (Phase 2 and the NEVER rules) — this file is the lookup table, not the contract.

Issue Solution
"no .github/workflows directory" Run from a repo root that contains GitHub Actions workflows
PyYAML missing pip install pyyaml (the scanner's only third-party dep)
gh not installed / not logged in The scan still runs; up to eight checks go unmeasured and are reported as such — the impostor-SHA vector, and the seven API-gated config facts (required-checks-skippable, fork-PR approval — when a CODEOWNERS rule covers workflows — whether the branch enforces it, the two environment checks, whether every secret lives in an environment, and whether the default branch requires a reviewed pull request). gh auth login to enable them; already logged in and still unmeasured usually means a setting only repository admins can read: classic branch protection (so "protected branches only" environment verdicts too, and the default-branch review check when no ruleset settles it), ruleset bypass lists, and the secret lists — re-run as an admin or check the setting on the repository's settings pages
Scanner emits zero findings Most likely the workflows are actually clean — that's the headline, not a bug. Otherwise check for a detector regression via tests/
Scanner exits non-zero or writes unparseable output Coverage failure, not a clean repo — surface exit code + stderr and stop (SKILL.md NEVER rules)
Subagent stops with a question Surface it to the user; the finding's heading stays unmarked until the subagent completes
run.py exits 2 with an argv complaint The --repo expansion was collapsed into one token by zsh — use the two-token ${REPO:+--repo} ${REPO:+"$REPO"} form from Phase 2
gh signed in, but required-checks-skippable, fork-PR approval and CODEOWNERS enforcement still unmeasured The origin remote is not a plain github.com URL (an ssh host alias such as git@github-work:, a port or www. in the URL, GitHub Enterprise Server, or no origin), so no owner/repo was derived. For a github.com repository, export CI_SECURE_REPO=owner/repo and re-run. GitHub Enterprise Server is not supported: those facts stay unmeasured
gh signed in, origin fine, but required-checks-skippable is unmeasured with "verdict assertion not proved (...)" The required check is a summary job whose pass/fail assertion the scanner could not prove; the supported forms are listed under Required-verdict coverage. Common fixes: set shell: bash explicitly on the summary step when the runner is not a standard GitHub-hosted ubuntu/macos label; pin re-actors/alls-green to the full commit SHA the scanner recognizes (b5b5b37504aa4183270bd3d855c52a67f212be35); or express the check as test "${{ ... }}" = "true" or a bash if [[ ... ]]; then exit 1; fi guard (double brackets, quoted operands) over needs.<job>.result

Deriving owner/repo from the remote

The Phase 2 block passes --repo owner/repo and derives it in the SAME shell that runs the scan: shell variables do not survive between an agent's separate commands, so a value set while following Phase 1 would be empty by Phase 2 and the lookups that need it would silently go unmeasured. Precedence: CI_SECURE_REPO, if the user exported it (to point the lookups at another repository, or to name one the remote does not); otherwise, only when gh is authenticated, the origin remote. A generic REPO in the environment is deliberately ignored: other tools and shell profiles set it, and a stale one would read another repository's settings and run history and score them as this one's. The origin arrives in two shapes:

# https://github.com/owner/repo[.git]  or  git@github.com:owner/repo.git
git remote get-url origin \
  | sed -nE 's#/$##; s#\.git$##; s#^(https://([^@/]+@)?github\.com/|(ssh://)?git@github\.com[:/])##p'

sed -n with the p flag prints only when the github.com prefix actually matched, so a remote without one (another host, or a relative path such as owner/repo.git) yields nothing. The value, from either source, is passed only if it is exactly owner/repo — one /, no scheme, no .git suffix — and anything else leaves REPO empty rather than passing it on: the scan still runs, and the checks it feeds are reported as skipped or unmeasured, never as passes.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warningtoday3 checks · Risk SAFE
  • Gen Agent Trust Hubtoday

    The skill 'ci-secure' is a comprehensive security auditing tool for GitHub Actions. It is authored by starslingdev and follows excellent security practices, including centralized data sanitization to prevent Markdown injection and defensive prompt engineering to protect against malicious instructions in audited code. Its use of dynamic loading and subprocess calls is restricted to essential functions for modularity and repository analysis using standard tools like git and gh.

  • Socket4d

    No alerts

  • Snyktoday

    Risk: MEDIUM · 2 issues

Signed by skilld at 9a84115. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 16 hours ago.

Activeupdated last week

README badge

README badge for starslingdev/skills/ci-secure